Semiparametric Estimation of First-Price Auctions with Risk-Averse Bidders

In view of the non-identification of the first-price auction model with risk-averse bidders, this paper proposes some parametric identifying restrictions and a semiparametric estimator for the risk aversion parameter(s) and the latent distribution of private values. Specifically, we exploit heterogeneity across auctioned objects to establish semiparametric identification under a conditional quantile restriction of the bidders' private value distribution and a parametrization of the bidders' utility function. We develop a multistep semiparametric method and we show that our semiparametric estimator of the utility function parameter(s) converges at the optimal rate, which is slower than the parametric one but independent of the dimension of the exogenous variables thereby avoiding the curse of dimensionality. We then consider various extensions including a binding reserve price, affiliation among private values, and asymmetric bidders. The method is illustrated on U.S. Forest Service timber sales, and bidders' risk neutrality is rejected.